CNC Interface and Torch Voltage Capture

[EmptyNester]

I started lookign at the CNC interface.

When the CNC/manual switch is in the top position (CNC/Track Torch 4T) the torch doesn't trigger with lines 1 and 2 shorting on the CNC connector. That only works when the switch is in the bottom position (Manual Torch 2T).

So, it works fine in the manual mode, but I wondered if there is something about the CNC mode that I'm missing?

I also did a capture of the voltage on the CNC output port. Here's what I got.



I captured the voltage on pins 4 and 6. I didn't change the voltage divide jumper, but with a peak voltage of a little over 12 volts, it looks like the jumper is already in the 1/16 position. The X axis shows the sample number. The samples are 20 ms. apart.

From the graph, I'm guessing that the intial full voltage is prior to the arc igniting. I would guess that the voltage drops on ignition and then recovers as the torch stabilizes. Does that seem to make sense?

[SeanMurphy265]

pins 1 and 2 are the relay
I have pins 5 and 7 wired for the full arc voltage

pins 4 and 6 should read the divided arc voltage.

Check 5 and 7 to see if they read the full arc voltage.

Which THC are you running?

[EmptyNester]

Thanks Sean.

I understand the pin out. I'm actually trying to understand the behavior of the voltage output so that I can write my own THC software. The graph shows the noise on the output with the divided voltage.

I do plan on looking at the noise on the undivided voltage to compare noise levels. However, I'd prefer to use the 1/16th voltage over raw voltage if the noise is comparable.

I still don't get what the front panel switch does though.

[Rambozo]

From what I understand, in 2T the arc is on whenever the switch lines connected, while in 4T the arc is started with one pulse then terminated with another. I don't have one to play with so I'm not sure if the actual switch line is brought high or low. I look forward to your experiments with an arduino THC. Because the voltage is an analog signal, I have always wondered if it would be simple to design an analog servo circuit to do THC. The noise levels are really something, I didn't expect that, but I guess it shouldn't surprise me with anything arcing.

[SeanMurphy265]

Which plasma cutter are you running?
If your running a multi purpose unit look at Zalkys thread at cnczone.com. From what I have read the multi-purpose units are the most noisy units for THC. The guy at CandCNC.com actually bought Zalky's unit to do some testing.

[everlastsupport]

I started lookign at the CNC interface.

When the CNC/manual switch is in the top position (CNC/Track Torch 4T) the torch doesn't trigger with lines 1 and 2 shorting on the CNC connector. That only works when the switch is in the bottom position (Manual Torch 2T).

So, it works fine in the manual mode, but I wondered if there is something about the CNC mode that I'm missing?

I also did a capture of the voltage on the CNC output port. Here's what I got.



I captured the voltage on pins 4 and 6. I didn't change the voltage divide jumper, but with a peak voltage of a little over 12 volts, it looks like the jumper is already in the 1/16 position. The X axis shows the sample number. The samples are 20 ms. apart.

From the graph, I'm guessing that the intial full voltage is prior to the arc igniting. I would guess that the voltage drops on ignition and then recovers as the torch stabilizes. Does that seem to make sense?

Some controllers flip a relay on and off (2T) , some pulse (4T), so the 2T and 4T can handle either.

On the graph, was that just firing the torch and pilot arc, but not cutting anything?

Most PIC and AVR chips have A/D if needed and you can add a cap and resister and time the charge. Never played with the pre-made (arduino) boards. But PICs are so cheap, internal OSC, low power, I would lean that way and all the free compilers.

[EmptyNester]

Which plasma cutter are you running?
If your running a multi purpose unit look at Zalkys thread at cnczone.com. From what I have read the multi-purpose units are the most noisy units for THC. The guy at CandCNC.com actually bought Zalky's unit to do some testing.

I'm using a PP50.

This was a quick/informal test. I didn't do anything with respect to shielding, ferrites, etc. I just wanted to see some waveforms.

[EmptyNester]

Some controllers flip a relay on and off (2T) , some pulse (4T), so the 2T and 4T can handle either.

On the graph, was that just firing the torch and pilot arc, but not cutting anything?

Most PIC and AVR chips have A/D if needed and you can add a cap and resister and time the charge. Never played with the pre-made (arduino) boards. But PICs are so cheap, internal OSC, low power, I would lean that way and all the free compilers.

Thanks for the info. I didn't know that 4T required pulsing. I'd rather stick with 2T so a cable break will kill the torch.

I have access to a lot of different embedded development platforms and tools. I figured Arduino would be good because it is:
- cheap processors
- has plenty of digital and analog I/O
- cheap prototype boards
- cheap displays (notice the theme)
- readily available
- well documented
- has a decent tool chain (though no debugger)
- has good support forums

I'm currently using a Mega256 board and a 2 line 16 character display.

I haven't yet done the voltage divider/filter/inverter yet. I wanted to decide what input voltage to use first (based on signal to noise ratio).

[everlastsupport]

Thanks for the info. I didn't know that 4T required pulsing. I'd rather stick with 2T so a cable break will kill the torch.

I have access to a lot of different embedded development platforms and tools. I figured Arduino would be good because it is:
- cheap processors
- has plenty of digital and analog I/O
- cheap prototype boards
- cheap displays (notice the theme)
- readily available
- well documented
- has a decent tool chain (though no debugger)
- has good support forums

I'm currently using a Mega256 board and a 2 line 16 character display.

I haven't yet done the voltage divider/filter/inverter yet. I wanted to decide what input voltage to use first (based on signal to noise ratio).

If it is 10 volts. Making the voltage divider easy. So the theme I take it was cheap. hahaha.

[EmptyNester]

Keeping with the prior theme - cheap I'm using one of the open source DSO201's (Digital Storage Scope 201). It is an iPod sized o-scopes that is about $70 on ebay. Its not a precision instruments, but "blunt instrument" seems fine for this exercise. (I haven't had a chance to compare it to a real scope.) And, its better for visualization than a meter.



I was using a divide by 10 probe. It looks like the noise at start-up (pilot arc only) is around 20 volts and dimishes to about 3 or 4 volts. So, from a filtering viewpoint - using the full voltage output instead of the divided output would be significantly better. (No real surprise there.)

The odd thing is that you don't see the voltage drop out on these graphes that were on the divided voltage graphs. I'm guessing that's a scope quality issue. I did these captures at 200 micro seconds and the divided voltage ones at 20 milliseconds

I haven't yet tried to cut steel while capturing the voltage as I'd need three hands to do that safetly.

[everlastsupport]

I have a couple nice storage scopes, but I carry a cheap one in the computer bag. http://www.seeedstudio.com/depot/dso...html?cPath=174

Nice unit for the price, I was shocked when I saw it work, really work.

[EmptyNester]

So, I used a 100K ohm and a 2.2K ohm for a voltage divider.

If I metered the raw output, it was around 140 volts. Based on the voltage divider it should read 3.014 on the o-scope. It looks like I'm getting about half of that. But I guess that's due to the resistors in the CNC interface.



I guess the thing I don't understand is that I thought the torch voltage is negative. But I'm reading a positive voltage. Is that because the signs are reversed on the CNC port pin-out description?

[EmptyNester]

Just found this article on how a THC works. It had a good listing of challenges.

http://www.fabricatingandmetalworkin...pplications-2/

The software itself shouldn't be hard to do, figuring out how it should work is the challenge. The article lists "Levels of Complexity". A brief summary is:
1) cutting speed changes: going slower gives a wider kerf and can cause torch dive
2) Gas pressure flucuations: pressure changes can cause the torch to go up or down
3) Kerf crossing: crossing a kerf can cause dive
4) Timing issues: is the torch is active at the end of the cut while the system decelerates the torch will dive
5) Electrode wear: as the electrode wears the tip will get closer to the metal

My initial thoughts on these are:

1) Not real sure about this one
2) I think this is best addressed by the compressor and regulator
3) kerf crossing is easy to adjust for by looking at the rate of change of voltage. I would expect kerf crossing to cause a large immediate jump in voltage versus warped metal giving a slower change in voltage
4) I would think this is handled by Mach
5) It would be easy to track tip changes and cutting time and have some compensation factor

Anyone have any other thoughts?

[Rambozo]

Good article. As to the kerf getting wider as the torch slows, it would be real cool to give the CNC control over the plasma power so that compensations could be made. Not sure about programming Mach 3 to do that, but I think it could be done. Possibly even the smoothing in software or hardware, to average the voltage, and remove noise, would take care of the kerf crossing. As to electrode wear, you could do a voltage re-calibrate when the torch measures the plate height. Once the cut starts, you will know the voltage at the desired height, so you would just take that as the following voltage.

[EmptyNester]

I ended up with a voltage divider of 100k and 2.53k. This gave me a divided voltage reading of 1.2V while the full output voltage was at 50V.

The voltage reading I used was a meter on the output. The difference between the meter and the scope is the scope is about 0.4 volts higher. I trust the meter more than the scope (scope is a DSO201 and I used two different meters). The graph below shows the full voltage with my divider and using a 100 mf filter capacitor.



I still don't understand why I'm not seeing a negative voltage on the output of the CNC port since the cutting voltage is negative.

I did connect both my scope and my meter between the CNC port outputs and the ground on an Arduino board (running on A/C). I check both the positive and negative torch voltage outputs (one, then the other) against the Arduino ground.

When powering up the Arduino, the meter reading jumped on very quickly (too fast to read), but then settled at around 84 milivolts. The scope traces for both are the similar.



I'm not sure what's going on. I'll talk to my consulting electric engineer tomorrow to see if he can shed any light on what I'm seeing. Any thoughts anyone else has would be appreciated.

[Rambozo]

I was under the impression that the torch outputs were basically isolated from ground, so that all measurements must be taken between them and not referenced to ground as far as cutting arc voltage is concerned?

[EmptyNester]

I was under the impression that the torch outputs were basically isolated from ground, so that all measurements must be taken between them and not referenced to ground as far as cutting arc voltage is concerned?

If that's the case, it might make the A/D circuit simpler. I'm under the impression that if the voltage is isolated from ground and its negative, you can just swap positive and negative and read it as a positive voltage.

Can anyone from Everlast weigh in on isolated grounds?

I wonder if the fact that I've only dealt with the pilot arc voltages so far figures in in any way too.

[EmptyNester]

I talked to my consulting electrical engineer. He agreed that it's probably an isolated supply. He thought that the a/c wave form I saw between grounds was probably just noise and so low a current it wouldn't be an issue.

I got the voltage divider set up. The calculated values weren't giving me the expected voltages. I believe that it's due to the existing resistors in the CNC interface. So, I went with the "scientific empirical testing model". I just kept trying different resistor combinations until I got the voltage I wanted

I also decided I'm going to add a couple zener diodes to protect the Arduino from overvoltage.

Once I get it all hooked up and know it works - I'll publish a schematic.

I'm not going to do anything with the torch height control loop yet. Since I don't have my table finished, it won't do me any good.

I may start working on some other aspects of the software (calibration, test mode, etc).

I haven't decided if I'd write a PC app to talk to it. I went to the trouble of isolating the power supplies for the BOB. If I do a serial connection and don't isolate the serial port, it would make the BOB isolation a wasted effort. I'll probaby wait until I know Mach better to understand if there's value in interfacing the Arduino with it via a serial port.

Just spent the day working on the gantry rail adjustment so that I can move the overall project along.

If the THC works, I'm curious what level of interest there would be in others duplicating it. If I can get it to work, I'll make all the info available, but I'm doing it specifically for the PP50. At this point I'm not real interested in making it a general solution.

[Rambozo]

Since you are working with Arduino for the THC, here is something interesting I stumbled across that might interest you.
As an alternative to Mach 3, a complete open source Arduino based CNC, GRBL. I can really see using this for some smaller projects that have been on the back burner for a long time.

http://dank.bengler.no/-/page/show/5470_grbl

http://www.shapeoko.com/wiki/index.php/Grbl

And a slick little driver board to go with it.
http://blog.makezine.com/arduino/grbl/

[everlastsupport]

Since you are working with Arduino for the THC, here is something interesting I stumbled across that might interest you.
As an alternative to Mach 3, a complete open source Arduino based CNC, GRBL. I can really see using this for some smaller projects that have been on the back burner for a long time.

http://dank.bengler.no/-/page/show/5470_grbl

http://www.shapeoko.com/wiki/index.php/Grbl

And a slick little driver board to go with it.
http://blog.makezine.com/arduino/grbl/

I have a mill on the way with servos, but this is a pretty slick controller. Wonder if there has been a servo hack for it.

Assuming small scale based on the spark fun motors? The Shapeoko mod one could add a double throw switch and turn it on and off, not sure if that would gain anything.

Ram, do you have any links to full blow system plans for the GRBL. Looks cheap to build. $70 controller, not sure what an Uno (is) or cost but my guess would be about $100 or so for both, then motors and start building.

Could be a fun projects for my son and I and we have spare laptops so maybe something portable. I have a PP50 on the way soon, waiting on it now.